A method for obstacle avoidance in narrow passages for multi-AUVs based on artificial potential field method

Abstract

This invention relates to the field of intelligent control technology for autonomous underwater vehicles (AUVs), specifically a method for obstacle avoidance in narrow passages using multiple AUVs based on an artificial potential field method. The method includes: determining whether an AUV swarm has entered a narrow passage; after the AUV swarm enters the narrow passage, calculating the minimum distances between the AUVs and obstacles, and between the AUVs themselves, based on an adaptive ocean current compensation mechanism, and dynamically calculating compensation factors based on a piecewise adaptive mapping function to reduce the collision rate of the AUVs; decomposing the resultant force of the AUVs using a virtual guidance method, decomposing the resultant force into components parallel and perpendicular to the guide line direction, suppressing the perpendicular component to eliminate the left-right swaying of the AUVs; and employing a path optimization method based on distance safety level assessment to smooth the jagged path generated by the AUVs during obstacle avoidance. This method enables the AUV swarm to autonomously avoid obstacles in narrow passages, ensuring that the AUV swarm can pass through narrow passages safely and efficiently in a certain formation.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for autonomous underwater vehicles, and in particular to a method for obstacle avoidance in narrow passages for multiple AUVs based on the artificial potential field method. Background Technology

[0002] In recent years, Autonomous Underwater Vehicles (AUVs) have developed rapidly in the field of marine equipment. Currently, the level of intelligence of individual AUVs has reached a high level. However, with the expansion of their application scenarios and the increasing demand for efficiency, multi-AUV collaborative operation is an important development direction. Multi-AUV systems can not only significantly improve operational efficiency and coverage, but also effectively cope with individual failures through task reallocation and functional redundancy, improving mission fault tolerance and reliability. However, due to the complex underwater environment, this requires AUVs to have a certain real-time obstacle avoidance capability. Unlike obstacle avoidance in open water, obstacle avoidance for multi-AUVs in narrow passages can lead to formation instability and low collaborative obstacle avoidance efficiency. A mature solution to this problem has not yet been found.

[0003] Research on cooperative obstacle avoidance algorithms is mainly divided into two categories: heuristic and learning-based. Early methods were mostly based on single-agent path planning and were extended to multi-agent systems, such as graph search methods like A*, RRT, and APF. With the increase in system scale and dynamics, Model Predictive Control (MPC) methods were later applied to obstacle avoidance. In recent years, with the rise of neural networks, learning-based algorithms for cooperative obstacle avoidance have emerged, such as MADDPG and PPO algorithms based on deep reinforcement learning. Currently, the Artificial Potential Field (APF) method is widely used in engineering due to its simple principle and clear physical meaning of parameters. However, research on multi-AUV obstacle avoidance methods for narrow passages is relatively limited; existing methods are all based on APF. Sun Hui et al. proposed an algorithm that integrates APF and model prediction for the typical environment of narrow passages. Model prediction uses the expected parameters provided by the APF algorithm to perform multi-step prediction and real-time optimization of the unmanned surface vessel's motion. Pan Wuwei et al. decomposed the virtual structure into a three-layer structure of "reference point-medium point-AUV" and then combined it with the APF algorithm to achieve collision avoidance. Bui et al. optimized APF obstacle avoidance from "static single potential field" to "dynamically adaptable hierarchical behavior", enabling robot swarms to adapt to obstacle avoidance needs in complex scenarios such as narrow spaces.

[0004] Cooperative obstacle avoidance algorithms are mainly divided into two categories: heuristic and learning-based. Heuristic obstacle avoidance methods mainly include graph search methods such as A* and RRT. These methods were initially applied to single-agent systems and later extended to multi-agent systems, but their computational cost increases exponentially with scale. As the scale and dynamics of systems increase, optimization methods based on Model Predictive Control (MPC) and the emerging deep reinforcement learning methods such as MADDPG and PPO have gradually become research hotspots. However, MPC-based algorithms frequently encounter deadlocks in dense scenarios; while obstacle avoidance algorithms based on deep reinforcement learning can balance optimality and adaptability, they rely on heavy computation or extensive training, placing high demands on onboard computing power, communication bandwidth, and sample data.

[0005] Currently, due to the "black box" nature of learning methods, engineering applications are mainly heuristic. Among them, the Artificial Potential Field (APF) method models "obstacle avoidance-cooperation" as a unified model of potential field gradient descent. It does not require explicit trajectory prediction or offline training, and can directly convert formation error into potential difference, avoiding the time lag caused by hierarchical planning. Therefore, current research on the scenario of narrow passages is based on this method.

[0006] Sun Hui et al. proposed an algorithm integrating APF (Automatic Collision Avoidance) and model prediction for the typical environment of narrow passages. Model prediction utilizes the expected parameters provided by the APF algorithm to perform multi-step prediction and real-time optimization of the unmanned surface vessel's (USV) motion. Pan Wuwei et al. decomposed the virtual structure into a three-layer structure of "reference point-medium point-AUV" and then combined it with the APF algorithm to achieve collision avoidance. Bui et al. optimized APF obstacle avoidance from a "static single potential field" to "dynamically adaptable layered behavior," enabling robot swarms to adapt to obstacle avoidance requirements in complex scenarios such as narrow spaces. However, the above methods are all adjustments made for the typical environment of "narrow passages" and do not consider whether the oscillation suppression processing is effective in other environments. Furthermore, the APF algorithm itself also suffers from path oscillation problems.

[0007] To address the aforementioned issues, Zhao Tianlong et al. improved the APF by designing an adaptive step-size adjustment strategy that incorporates distance-related gravity coefficients, repulsion coefficients, and obstacle influence weights. However, this method was only validated under static cylindrical obstacles and did not consider other complex scenarios. Bian et al. simplified the path to reduce unnecessary turns by using a redundant node removal strategy and inserted auxiliary nodes at path turning points based on the obstacle repulsion potential field to avoid path distortion caused by smoothing. However, this method did not consider the smooth connection of adjacent target segments, and the fixed weights may lead to abrupt changes in the turning angle.

[0008] For underwater environments, Cao et al. took into account the influence of ocean currents and integrated velocity synthesis (VS) and confidence function (BF) on the basis of the APF algorithm, so that AUVs could offset the influence of ocean currents through the VS method. However, the parameters for ocean current compensation are fixed values, which will result in a high collision rate in practice. In addition, the improved BF introduced a dead point function, which made the overall algorithm more complex and unable to respond to environmental changes in a timely manner, thus weakening the real-time performance and effectiveness of path planning. Summary of the Invention

[0009] The purpose of this invention is to overcome the above-mentioned defects in the existing technology and propose a multi-AUV obstacle avoidance method in narrow passages based on the artificial potential field method. This method enables AUV clusters to autonomously avoid obstacles in narrow passages, ensuring that AUV clusters can pass through narrow passages safely and efficiently in a certain formation.

[0010] The technical solution of this invention is: a method for obstacle avoidance in narrow passages for multiple AUVs based on the artificial potential field method, comprising the following steps:

[0011] S1. Determine if the AUV group has entered a narrow passage;

[0012] S2. After the AUV cluster enters the narrow channel, the minimum distance between the AUV and the obstacle, as well as between the AUVs, is calculated according to the adaptive ocean current compensation mechanism. The compensation factor is dynamically calculated based on the piecewise adaptive mapping function to reduce the collision rate of the AUVs.

[0013] S3. Use the virtual guidance method to decompose the resultant force of the AUV into components that are parallel to and perpendicular to the direction of the guide line, and suppress the vertical component to eliminate the left and right sway of the AUV.

[0014] S4. A path optimization method based on distance safety level assessment is adopted to smooth the jagged path generated by the AUV during obstacle avoidance.

[0015] In this invention, step S1 includes the following specific implementation steps:

[0016] S1.1 Determine if the AUV cluster has reached the target point:

[0017] If all AUVs stop sailing upon reaching the target point, the mission ends; if the target point is not reached, each AUV will monitor the obstacle position information measured by the forward-looking sonar in real time. This information is stored in the database, which is set to store the position information collected by the forward-looking sonar in the last 4 seconds.

[0018] S1.2 Obstacle Monitoring:

[0019] When there is no obstacle information, the database is empty; when an obstacle appears, the data in the database is updated, and the information in the database is represented as follows:

[0020] ,

[0021] ,

[0022] Taking the direction of AUV travel as positive, where, This represents the set of obstacle locations detected by the AUV on the left. This represents the set of obstacle locations detected by the AUV on the right. This represents a set of 4 consecutive seconds. Indicates the obstacle on the left. Position coordinates at that moment Indicates the obstacle on the right. Position coordinates at that moment;

[0023] S1.3. Based on obstacle monitoring, determine whether the AUV convoy has entered a narrow passage and adjust the AUV formation accordingly:

[0024] When the database is not empty, select the obstacle location points closest to the AUV from the window array. and Based on both calculations, the forward environmental width is:

[0025] ,

[0026] The formation of the AUV cluster is adjusted according to the ambient width: when the ambient width is less than the minimum safety margin, the AUV cluster cannot pass through a narrow passage and pauses; when the ambient width is less than or equal to the minimum formation width required to maintain the original formation, but greater than the minimum safety margin, the AUV cluster passes through the passage in a straight line; when the ambient width is less than or equal to the current formation width but greater than the minimum formation width required to maintain the original formation, the AUV cluster reduces its width while maintaining the original formation shape; when the ambient width is greater than the current formation width, the original formation remains unchanged.

[0027] AUV formation is represented as follows:

[0028] ,

[0029] in, This indicates a horizontal, linear formation. Indicates the initial formation; Minimum safety margin; Indicates configuration parameters; The turning radius of the AUV; This represents the reduction factor of the formation. ; This indicates the minimum formation width required to maintain the original formation. Indicates the current formation width.

[0030] In step S2, the adaptive ocean current compensation mechanism utilizes a dual distance sensing strategy;

[0031] No. AUV at time Shortest Euclidean distance to all obstacles for:

[0032] ,

[0033] in, Indicates the first The current location of each AUV. Represents a set of obstacles. This indicates the number of obstacles among all obstacles within the range of the AUV's forward-looking sonar scan. The position vectors of the obstacles;

[0034] No. AUV at time Shortest Euclidean distance to other AUVs for:

[0035] ,

[0036] Obstacle compensation factor for the distance between the AUV and the obstacle. for:

[0037] ,

[0038] in, This represents the warning radius, which is the critical distance at which the AUV detects an obstacle ahead and triggers the corresponding compensation factor.

[0039] AUV compensation factor for the distance between AUVs for:

[0040] ,

[0041] Let the first Comprehensive compensation factor for each AUV This represents the minimum value of the obstacle compensation factor and the AUV compensation factor at the same moment, i.e.:

[0042] ,

[0043] Limit the rate of change of the comprehensive compensation factor and set the first... Smoothing compensation factor for each AUV for:

[0044] ,

[0045] in, This represents the smoothing compensation factor from the previous time step. This represents the maximum permissible change in the compensation factor per time step. Indicates a temporary constraint value;

[0046] No. The base speed of an AUV This velocity is the vector sum of the velocities corresponding to each component force:

[0047] ,

[0048] in, Indicates the first AUV at time The velocity vector generated by the force at the target point; Indicates the first AUV at time The velocity generated by the repulsive force of an obstacle; Shown in AUV at time The velocity vector generated by the repulsive force between AUVs; Indicates the first AUV at time The velocity vector generated by the formation-maintaining force;

[0049] Based on the base velocity, ocean current compensation is applied to obtain the first... AUV at time Speed ​​after ocean current compensation for:

[0050] ,

[0051] in, The ocean current compensation term is represented by the following formula:

[0052] ,

[0053] in, This represents the velocity vector of the ocean current.

[0054] In step S3, the ocean current-compensated velocity obtained in step S2 is... Decompose the velocity into vertical and horizontal velocities:

[0055] ,

[0056] ,

[0057] in, This indicates the velocity parallel to the direction of the virtual guide line. This indicates the velocity perpendicular to the direction of the virtual guide line. A unit vector representing the direction of the AUV virtual guide line;

[0058] Then the parallel components are smoothed, and the smoothed parallel components are... for:

[0059] ,

[0060] in, Indicates the parallel smoothing coefficient;

[0061] The vertical component is suppressed, and the suppressed vertical component is:

[0062] ,

[0063] in, Indicates the vertical inhibition coefficient;

[0064] The suppressed velocity vector is then obtained as follows:

[0065] .

[0066] In step S4, the change in velocity angle is defined. for:

[0067] ,

[0068] in, This indicates that the input will be restricted to... interval;

[0069] Regarding environmental impact factors, the first consideration is the distance from the AUV to the obstacle. and warning distance Define security level :

[0070] ,

[0071] Then, the smoothing weight is calculated based on the change in velocity angle and the safety level. for:

[0072] ,

[0073] The final output is the desired velocity vector. for:

[0074] ,

[0075] The desired velocity vector obtained Converted into controller input ,

[0076]

[0077] in, Indicates the current AUV The magnitude and direction of the forward movement at any given moment;

[0078] right Apply saturation constraints, and then apply the saturation constraints to the input. As the input signal for AUV, The magnitude and direction of the AUV's acceleration at the next moment:

[0079] ,

[0080] in, This indicates the maximum limit value of the input signal.

[0081] The beneficial effects of this invention are:

[0082] (1) This application proposes an adaptive ocean current compensation mechanism. The mechanism adopts a dual distance perception strategy to calculate the minimum distance between the AUV and the obstacle and the AUV, respectively, and dynamically calculates the compensation factor based on the piecewise adaptive mapping function. The compensation intensity is adaptively adjusted by environmental perception, which reduces the collision rate of the traditional fixed compensation method in the complex marine environment and improves the formation accuracy.

[0083] (2) This application uses the virtual guidance method to decompose the resultant force into components parallel and perpendicular to the direction of the guide line, suppressing the vertical velocity component and reducing the oscillation caused by the channel itself acting as an obstacle in the narrow channel, so as to eliminate the left and right swing of the AUV.

[0084] (3) This application proposes a path optimization method based on distance safety level assessment. By comprehensively considering environmental factors and speed angle changes, it smooths the sawtooth path caused by the frequent alternation of forces due to obstacle avoidance turning in narrow channels by the APF algorithm. Under the premise of ensuring obstacle avoidance safety, it effectively suppresses the sawtooth path and reduces oscillation.

[0085] (4) By using the method described in this application, the speed and direction of movement of each AUV in the AUV cluster at the next moment are determined, and the speed and direction of movement of each AUV in the AUV cluster at each moment when passing through the narrow passage are accurately indicated, so as to realize the optimal planning of the movement trajectory of the AUV cluster when passing through the narrow passage and ensure that the AUV cluster can pass through the narrow passage smoothly.

[0086] In summary, after multiple AUVs enter a narrow passage, they can avoid and traverse the passage by means of an adaptive ocean current compensation mechanism, a virtual guidance method, and a safety level assessment method. The three methods work together to ensure that the AUV cluster can safely and efficiently pass through the narrow passage. Attached Figure Description

[0087] Figure 1 This is a flowchart of the method described in this application;

[0088] Figure 2 This is a schematic diagram of the resultant force decomposition based on the virtual guidance method;

[0089] Figure 3 This is a schematic diagram of a path optimization method based on distance safety level assessment;

[0090] Figure 4 This is a simulation diagram of multiple AUVs passing through a narrow passage before the improvement;

[0091] Figure 5 yes Figure 4 A magnified view of a portion of the image;

[0092] Figure 6 This is a simulation diagram of the improved multi-AUV passing through a narrow passage;

[0093] Figure 7 yes Figure 6 A magnified view of a portion of the image;

[0094] Figure 8 This is a simulation diagram of multiple AUVs passing through a narrow passage using the method described in this application after the formation width is reduced. Detailed Implementation

[0095] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0096] Specific details are set forth in the following description to provide a full understanding of the invention. However, the invention can be practiced in many ways other than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0097] This invention discloses a multi-AUV obstacle avoidance method for narrow passages based on the artificial potential field method. AUV swarm collaboration typically employs centralized or distributed information exchange methods. This application proposes a distributed method for AUV swarm information exchange, where each AUV in the swarm can autonomously react and adjust its state based on environmental information perceived by its own sensors and information sent by other AUVs, ultimately completing the task. Therefore, when an AUV swarm encounters a narrow passage, the AUVs can communicate passage information through this method to achieve overall passage avoidance and passage. The overall process is as follows: Figure 1 As shown, the method specifically includes the following steps.

[0098] The first step is to determine whether the AUV cluster has entered a narrow passage.

[0099] First, it determines whether the AUV cluster has reached the target point, i.e., whether the mission has been completed. If the mission is completed, all AUVs stop traveling, and the mission ends. If the target point has not been reached, each AUV will monitor the obstacle position information measured by the forward-looking sonar in real time. This information is stored in a database, which is set to store the position information collected by the forward-looking sonar within the most recent 4 seconds. When obstacle information is present in the database for 4 consecutive seconds, it indicates that the AUV has entered a narrow passage.

[0100] Then, obstacle detection is performed. When there is no obstacle information, the database is empty; when an obstacle appears, the data in the database is updated, so the information in the database is represented as follows:

[0101] ,

[0102] ,

[0103] Taking the direction of AUV travel as positive, where, This represents the set of obstacle locations detected by the AUV on the left. This represents the set of obstacle locations detected by the AUV on the right. This represents a set of 4 consecutive seconds. Indicates the obstacle on the left. Position coordinates at that moment Indicates the obstacle on the right. The position coordinates at that moment.

[0104] Subsequently, based on obstacle monitoring, it was determined whether the AUV convoy had entered a narrow passage, and the AUV formation was adjusted accordingly.

[0105] When the database is not empty, select the obstacle location points closest to the AUV from the window array. and Based on both calculations, the forward environmental width is:

[0106] ,

[0107] The formation of the AUV swarm is adjusted based on the ambient width: when the ambient width is less than the minimum safety margin, the AUV swarm cannot pass through narrow passages and pauses; when the ambient width is less than or equal to the minimum swarm width required to maintain the original formation, but greater than the minimum safety margin, the AUV swarm passes through the passage in a straight line; when the ambient width is less than or equal to the current swarm width but greater than the minimum swarm width required to maintain the original formation, the AUV swarm reduces its width while maintaining the original formation shape; when the ambient width is greater than the current swarm width, the original formation remains unchanged. In summary, the AUV swarm formation is represented as:

[0108] ,

[0109] in, This indicates a horizontal, linear formation. Indicates the initial formation; Minimum safety margin; Indicates configuration parameters; The turning radius of the AUV; This represents the reduction factor of the formation. ; This indicates the minimum formation width required to maintain the original formation. Indicates the current formation width.

[0110] The second step involves entering the narrow passage with the AUV cluster. Based on the adaptive ocean current compensation mechanism, the minimum distances between the AUVs and obstacles, as well as between the AUVs themselves, are calculated. The compensation factor is then dynamically calculated based on the piecewise adaptive mapping function to reduce the collision rate of the AUVs.

[0111] An adaptive ocean current compensation mechanism is proposed using a dual distance sensing strategy, with distance parameters including the distance between AUV and obstacles and the distance between AUVs.

[0112] No. AUV at time Shortest Euclidean distance to all obstacles for:

[0113] ,

[0114] in, Indicates the first The current location of each AUV. Represents a set of obstacles. This indicates the number of obstacles among all obstacles within the range of the AUV's forward-looking sonar scan. The position vectors of the obstacles.

[0115] No. AUV at time Shortest Euclidean distance to other AUVs (j≠i) for:

[0116] ,

[0117] After obtaining the two distance values, calculate the compensation factor for each distance value. Obstacle compensation factor for the distance between the AUV and the obstacle. Set to:

[0118] ,

[0119] in, This represents the warning radius, which is the critical distance at which the AUV senses an obstacle ahead and triggers the corresponding compensation factor.

[0120] Similarly, the AUV compensation factor for the distance between AUVs. Set to:

[0121] ,

[0122] Calculate the first Comprehensive compensation factor for each AUV This represents the minimum value of the obstacle compensation factor and the AUV compensation factor at the same moment, i.e.:

[0123] ,

[0124] To avoid abrupt changes in the comprehensive compensation factor, a limit is placed on the rate of change of the comprehensive compensation factor, setting the first... Smoothing compensation factor for each AUV for:

[0125] ,

[0126] in, This represents the smoothing compensation factor from the previous time step. This represents the maximum permissible change in the compensation factor per time step. This represents a temporary constraint value.

[0127] Therefore, we obtain the first... AUV at time The base velocity is generated by the combined forces of the target point force, obstacle repulsion, inter-AUV repulsion, and formation maintaining force. This velocity is the vector sum of the velocities corresponding to each component force:

[0128] ,

[0129] in, Indicates the first AUV at time The velocity vector generated by the force at the target point; Indicates the first AUV at time The velocity generated by the repulsive force of an obstacle; Shown in AUV at time The velocity vector generated by the repulsive force between AUVs; Indicates the first AUV at time The velocity vector generated by the formation holding force.

[0130] Based on the base velocity, ocean current compensation is applied to obtain the first... AUV at time Speed ​​after ocean current compensation for:

[0131] ,

[0132] in, The ocean current compensation term is represented by the following formula:

[0133] ,

[0134] in, This represents the velocity vector of the ocean current.

[0135] The third step involves the narrow passage itself acting as an obstacle, causing oscillations. The virtual guidance method is used to decompose the resultant force on the AUV, breaking it down into components parallel and perpendicular to the guide line direction. The perpendicular component is suppressed to eliminate the left-right swaying of the AUV.

[0136] As shown in Figure 2, the virtual guide line is the globally desired route generated by the follower based on the navigator's desired route, with the speed compensated for by the aforementioned ocean currents. Decompose the velocity into vertical and horizontal velocities:

[0137] ,

[0138] ,

[0139] in, This indicates the velocity parallel to the direction of the virtual guide line. This indicates the velocity perpendicular to the direction of the virtual guide line. The unit vector representing the direction of the AUV virtual guide line.

[0140] Then the parallel components are smoothed, and the smoothed parallel components are... for:

[0141] ,

[0142] in, This represents the parallel smoothing coefficient.

[0143] The vertical component is suppressed, and the suppressed vertical component is:

[0144] ,

[0145] in, This represents the vertical inhibition coefficient.

[0146] The suppressed velocity vector is then obtained as follows:

[0147] .

[0148] In the fourth step, the APF algorithm addresses the frequent alternation of attraction and repulsion caused by the AUV's obstacle avoidance and turning in narrow channels, resulting in a sawtooth path. A path optimization method based on distance safety level assessment is used to smooth the path. Taking into account environmental factors and speed angle changes, the sawtooth path is effectively suppressed and oscillations are reduced while ensuring obstacle avoidance safety.

[0149] The optimization method based on distance safety level assessment adopts an adaptive adjustment strategy by evaluating the generated velocity angle change and environmental impact, as shown in Figure 3. The change in velocity angle is defined. for:

[0150] ,

[0151] in, This indicates that the input will be restricted to... Interval.

[0152] Regarding environmental impact factors, the first consideration is the distance from the AUV to the obstacle. and warning distance Define security level :

[0153] ,

[0154] Then, the smoothing weight is calculated based on the change in velocity angle and the safety level. for:

[0155] ,

[0156] The final output is the desired velocity vector. for:

[0157] .

[0158] At this point, the desired velocity vector will be obtained. Converted into controller input :

[0159]

[0160] in, Indicates the current AUV The speed and direction of movement at any given moment.

[0161] right Apply saturation constraints and use the saturated input as the input signal for the AUV:

[0162] ,

[0163] in, This indicates the maximum limit value of the input signal.

[0164] The method described in this application determines the speed and direction of each AUV in the AUV cluster at the next moment by determining the magnitude and direction of acceleration of the AUV in the next moment. This method enables the indication of the speed and direction of each AUV in the cluster at every moment as it passes through a narrow passage, achieving optimal planning of the AUV cluster's trajectory when passing through the narrow passage and ensuring that the AUV cluster can pass through the narrow passage smoothly.

[0165] To verify the effectiveness of the method described in this application, obstacle avoidance simulation was conducted on multiple AUVs navigating through a narrow passage, and the results were compared with those before the improvement. The experiment involved five AUVs traveling in a pentagonal formation through an extremely narrow passage, with the five AUVs passing in a straight line. Figure 4 To improve the previous path diagram of each AUV through the channel. Figure 5 for Figure 4 A magnified view of a portion of the image; Figure 6 This is a path diagram of each AUV passing through the channel using the method of the present invention. Figure 7 yes Figure 6 A magnified view of a section. Comparison. Figure 5 and Figure 7 It can be seen that the method proposed in this invention significantly reduces path oscillation. Then, a relatively narrow channel was set up, and the formation shape was kept unchanged while passing through the channel, thus reducing the formation width. Formation simulation was performed, and the results show that the method proposed in this application can still successfully pass through narrow channels. The simulation results are as follows: Figure 8 As shown.

[0166] Because the method proposed in this invention incorporates an adaptive ocean current compensation mechanism, it significantly reduces the collision rate when passing through narrow passages. Specific test data is shown in Table 1. The collision rate of the improved obstacle avoidance method for narrow passages in this invention is much lower than that of existing obstacle avoidance methods.

[0167] Table 1 Collision Rate Test Results

[0168] algorithm Number of successes Obstacle avoidance success rate Collision rate Original Algorithm 8 / 30 26.67% 73.33% Improved method 27 / 30 90.00% 10.00%

[0169] The above provides a detailed description of the obstacle avoidance method for multiple AUVs in narrow passages based on the artificial potential field method provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make various improvements and modifications to this invention without departing from the principles of this invention, and these improvements and modifications also fall within the protection scope of the claims of this invention. The above description of the disclosed embodiments enables those skilled in the art to implement or use this invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of this invention. Therefore, this invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-AUV obstacle avoidance method in a narrow channel based on a potential field method, characterized in that, The method comprises the following steps: S1, judging whether the AUV cluster travels to a narrow channel or not; S2, after the AUV cluster enters the narrow channel, the minimum distances between the AUVs and obstacles and between the AUVs are calculated respectively according to an adaptive ocean current compensation mechanism, and a compensation factor is dynamically calculated based on a segmented adaptive mapping function to reduce the collision rate of the AUVs; S3, a virtual guidance method is used to decompose the resultant force of the AUVs into components parallel and perpendicular to the direction of the guidance line, and the perpendicular component is suppressed to eliminate the left and right swing of the AUVs; S4, a path optimization method based on distance safety level evaluation is used to smooth the zigzag path generated by the AUVs in the obstacle avoidance process; In step S2, the adaptive ocean current compensation mechanism uses a double distance perception strategy, The first AUV is at time The shortest Euclidean distance from the first AUV to all obstacles is at time t0. ， wherein, represents the current position of the th AUV, represents a set of obstacles, represents the position vector of the th obstacle among all obstacles within the forward-looking sonar scanning range of the AUV. No. AUV at time Shortest Euclidean distance to other AUVs for: ， Obstacle compensation factor for distance between AUV and obstacle is: ， wherein, represents the alert radius, i.e. the critical distance at which the AUV perceives the presence of an obstacle in front and triggers the corresponding compensation factor; AUV compensation factor for AUV-to-AUV distance is: ， Let the comprehensive compensation factor of the AUV be The minimum value of the obstacle compensation factor and the AUV compensation factor at the same time, that is:​ ， The rate of change of the comprehensive compensation factor is limited, and the first AUV smooth compensation factor is set as: ， wherein denotes the smoothing compensation factor of the last time step, denotes the maximum allowed variation of the compensation factor per time step, denotes a temporary constraint value; The first The base velocity of the AUV which is the vector sum of the component velocities. ， wherein, represents the velocity vector of the nth AUV at time generated by the target point force; represents the velocity of the nth AUV at time generated by the obstacle repulsion force; represents the velocity vector of the nth AUV at time generated by the inter-AUV repulsion force; represents the velocity vector of the nth AUV at time generated by the formation maintenance force; ​​​​ On the basis of the basic speed, the current compensation is carried out, and the first AUV speed after the current compensation is obtained at time is: ， wherein represents the ocean current compensation term, which is calculated as ， wherein represents the velocity vector of the ocean current.

2. The multi-AUV narrow channel obstacle avoidance method based on artificial potential field method according to claim 1, characterized in that, Step S1 comprises the following specific implementation steps: S1.1, judging whether the AUV cluster reaches the target point or not: If all the AUVs stop sailing after reaching the target point, the task is completed; If the target point is not reached, each AUV will monitor the obstacle position information measured by the forward-looking sonar in real time, and the information is stored in the database. The position information collected by the forward-looking sonar in the last 4 seconds is stored in the database. S1.2, obstacle monitoring: When there is no obstacle information, the database is empty; when an obstacle appears, the data in the database is updated, and the information in the database is represented as: ， ， with the AUV's forward direction being positive, denotes the set of obstacle positions detected by the left AUV, denotes the set of obstacle positions detected by the right AUV, denotes the set of consecutive 4 second time instances, denotes the position coordinates of the left obstacle at time instance t, denotes the position coordinates of the right obstacle at time instance t. S1.3, judging whether the AUV cluster travels to a narrow channel or not according to the obstacle monitoring, and adjusting the AUV formation shape: When the database is not empty, the closest obstacle position point to the AUV is selected from the window array respectively and The environmental width of the advance is calculated according to the two ， The formation shape of the AUV cluster is adjusted according to the width of the environment: when the width of the environment is less than the minimum safety margin, the AUV cluster cannot pass through the narrow channel, and the AUV cluster is paused; when the width of the environment is less than or equal to the minimum formation width required to maintain the original formation shape, but greater than the minimum safety margin, the AUV cluster passes through the channel in a linear formation; when the width of the environment is less than or equal to the current formation width which is greater than the minimum formation width required to maintain the original formation shape, the AUV cluster reduces the width on the basis of maintaining the original formation shape; when the width of the environment is greater than the current formation width, the original formation shape is maintained; The AUV formation shape is represented as: ， wherein, represents a lateral line formation; represents an initial formation; minimum safety margin; represents a configuration parameter; is a turning radius of the AUV; represents a reduction factor of the formation, ; represents a minimum formation width to maintain the original formation; represents a current formation width.

3. The multi-AUV narrow channel obstacle avoidance method based on artificial potential field method according to claim 1, characterized in that, In step S3, The current-compensated velocity obtained in step S2 is decomposed into a vertical direction velocity and a horizontal direction velocity. The current-compensated velocity obtained in step S2 is decomposed into a vertical direction velocity and a horizontal direction velocity. ， ， wherein, Vp represents the velocity parallel to the virtual guidance line direction, Vn represents the velocity perpendicular to the virtual guidance line direction, V represents the unit vector of the virtual guidance line direction of the AUV; The parallel component is then smoothed, and the smoothed parallel component is: ， wherein represents a parallel smoothing coefficient; The vertical component after suppression is: ， wherein represents the vertical suppression factor; Then the suppressed speed vector is obtained as: 。 4. The multi-AUV narrow channel obstacle avoidance method based on artificial potential field method according to claim 1, characterized in that, In step S4, Definition of the amount of change in the angle of velocity is: ， wherein, represents to restrict the input to an interval; For the environmental impact factor, first according to the distance of AUV to the obstacle and the alert distance define the safety level : ， Then, the smoothing weight is calculated according to the variation of the speed angle and the safety level is: ， The final output desired velocity vector is: ， The resulting desired velocity vector is converted to an input to the controller , wherein, represents the speed size and direction of the AUV advancing at the current time instant; To saturated, the saturated input as the input signal of the AUV to guide the size and direction of the AUV's acceleration at the next moment: ， wherein represents the maximum limit value of the input signal.

Images